From 389a3eaf89c53131fd4119575c4cd11492c7e4cd Mon Sep 17 00:00:00 2001 From: Noah Smith Date: Fri, 31 May 2024 10:52:22 +0200 Subject: [PATCH] starid89 --- _drafts/starid86-hubble.md | 25 ------------------------- _drafts/starid89-sirtf.md | 18 ++++++++++++++++++ _posts/2024-05-31-starid89-sirtf.md | 18 ++++++++++++++++++ 3 files changed, 36 insertions(+), 25 deletions(-) delete mode 100644 _drafts/starid86-hubble.md create mode 100644 _drafts/starid89-sirtf.md create mode 100644 _posts/2024-05-31-starid89-sirtf.md diff --git a/_drafts/starid86-hubble.md b/_drafts/starid86-hubble.md deleted file mode 100644 index f16ea55..0000000 --- a/_drafts/starid86-hubble.md +++ /dev/null @@ -1,25 +0,0 @@ ---- -layout: post -title: "starid86 hubble" -categories: aerospace starid ---- -[paper 1986](https://statespace.dev/docs/papers/1986%20groth.pdf){:target="_blank" rel="noopener"} - -the hubble space telescope played an important role in project starid, and this paper is part of that story. project starid began in 1990 as part of work with the hubble space telescope astrometry team, and the related texas minor planet project at utaustin. part of that work was based explicitly on the starid algorithm in this paper, published four years earlier by a member of the hst program on the east coast. - -as a result, this paper has special significance as one of the earliest elements of project starid. and because it captures the astronomical influences in a concentrated form, thirteen years before serious aerospace influences entered the story. instead of an isolated discussion of the paper, it's possible to explore exactly how the algorithm was used for hst astrometry and the tmpp, and in the context of the beginnings of project starid. - -the starting point for the normal workflow was a photographic glass plate taken with the cass-cam on the eighty-two inch telescope at mcdonald observatory. a plate was roughly the size and shape of a piece of writing paper, as thick as window glass, and stored in a white paper sleeve with a printed form on the front for observer's notes, filled out by hand. - -the dark patches of emulsion on the plate, formed by bright stars, were digitized first. the pds microdensitometer was used to raster scan these shapes, resulting in a set of digital images, one for each scanned star. all of the images were measured in the pds x, y coordinate frame. there was no point in scanning regions of the plate between stars. only bright stars were scanned, all in the global pds coordinate frame. the second step was to centroid each image, resulting in a list of x1, y1 pds star positions. - -one of the observer's notes on the plate sleeve was the position on the sky of the guide star used for telescope tracking during the plate exposure, allowing a list of nearby bright star positions to be extracted from a star catalog, this is the exact opposite of the lost in space problem, and instead of having to search the entire sky for stars matching those on the photographic plate, a list of x2, y2 catalog star positions was immediately available. - -now there were two lists of star positions, in different coordinate frames and with different numbers and orderings of stars. the challenge was to match some of the stars between the two lists. this was a common task for astronomers, performed manually throughout the decades of reliance on photographic plates. the new element becoming practical with eighties minicomputers and workstations was automation of the matching process. - -the first step of the paper's algorithm is to make the lists roughly twenty stars long, picking the brightest first. exact similarity of the two lists is not practical, but ideally they now contain approximately the same stars and the major differences between them are ordering and coordinate frames. the objective is to match at least some, and preferably most, of the stars between the two lists. with two or more matches, a coordinate frame transformation between the two lists can be estimated. - -now the same operations are performed on both lists. triangles are formed for all possible triplets of stars in a list. two features represent each triangle - the ratio of the shortest to the longest side, and the angle between them. both features are independent of coordinate frame, they don't depend on scale or orientation. - -direct matches can now be made between the lists of triangles with similar ratios and angles. if a star from the photographic plate list appears in many matching triangles, its matching star in the catalog list can be determined by set operations - union, intersection, etc. numbers of matching triangles essentially 'cast votes' on star matches. if many triangle match votes are cast for a star, then it's been identified. - diff --git a/_drafts/starid89-sirtf.md b/_drafts/starid89-sirtf.md new file mode 100644 index 0000000..791732d --- /dev/null +++ b/_drafts/starid89-sirtf.md @@ -0,0 +1,18 @@ +--- +layout: post +title: "starid89 sirtf" +categories: aerospace starid +--- +[paper 1986](https://statespace.dev/docs/papers/1989%20bezooijen.pdf){:target="_blank" rel="noopener"} + +in the seventies and eighties, nasa initiated two high-level scientific programs, both of which have played central roles in the project starid story - the [great observatories program](https://en.wikipedia.org/wiki/Harlan_James_Smith) looking outwards, and the [earth observing system](https://en.wikipedia.org/wiki/Earth_Observing_System) looking inwards. + +the hubble space telescope directly entered the starid story first, in 1990. it was the flagship of the great observatories program and well-known and loved by school-kids of the eighties. it's sibling, the [space infrared telescope](https://en.wikipedia.org/wiki/Spitzer_Space_Telescope) was less well-known, but important. + +this paper summarizes the star identification work done in the eighties for the sirtf attitude control system. the algorithm was also extended to a commercial star tracker, where the scope grew to include the lost in space problem. the result seems to be that the 'hi-fi' end of the star identification spectrum was targeted. in particular, both star brightness and position measurement accuracies play a prominent role. the direct relevance for the 'lo-fi' end of the spectrum is questionable. + +the indirect relevance is strong however, as this paper seems to be first where two triangles with a shared side are diagrammed and discussed directly. an intuitive diagram of the situation has four stars as the four corners of a shape like a children's kite, with the two support sticks of the kite dividing the shape into four quadrants. this paper seems to be the first appearance of this type of diagram. the two support sticks are the shared sides of two overlapping and complementary pairs of triangles. + +in this case, the shared sides are essentially used as secondary validation checks after the primary star identification algorithm. they aren't actively used as part of determining star identities, but are instead used to confirm that identities are valid. the primary determination of star identities uses methods similar to earlier papers, based on non-iterative direct matching of accurate star pair angles and brightnesses. the 'shared side validation' is an additional operation added on top of that. + + diff --git a/_posts/2024-05-31-starid89-sirtf.md b/_posts/2024-05-31-starid89-sirtf.md new file mode 100644 index 0000000..791732d --- /dev/null +++ b/_posts/2024-05-31-starid89-sirtf.md @@ -0,0 +1,18 @@ +--- +layout: post +title: "starid89 sirtf" +categories: aerospace starid +--- +[paper 1986](https://statespace.dev/docs/papers/1989%20bezooijen.pdf){:target="_blank" rel="noopener"} + +in the seventies and eighties, nasa initiated two high-level scientific programs, both of which have played central roles in the project starid story - the [great observatories program](https://en.wikipedia.org/wiki/Harlan_James_Smith) looking outwards, and the [earth observing system](https://en.wikipedia.org/wiki/Earth_Observing_System) looking inwards. + +the hubble space telescope directly entered the starid story first, in 1990. it was the flagship of the great observatories program and well-known and loved by school-kids of the eighties. it's sibling, the [space infrared telescope](https://en.wikipedia.org/wiki/Spitzer_Space_Telescope) was less well-known, but important. + +this paper summarizes the star identification work done in the eighties for the sirtf attitude control system. the algorithm was also extended to a commercial star tracker, where the scope grew to include the lost in space problem. the result seems to be that the 'hi-fi' end of the star identification spectrum was targeted. in particular, both star brightness and position measurement accuracies play a prominent role. the direct relevance for the 'lo-fi' end of the spectrum is questionable. + +the indirect relevance is strong however, as this paper seems to be first where two triangles with a shared side are diagrammed and discussed directly. an intuitive diagram of the situation has four stars as the four corners of a shape like a children's kite, with the two support sticks of the kite dividing the shape into four quadrants. this paper seems to be the first appearance of this type of diagram. the two support sticks are the shared sides of two overlapping and complementary pairs of triangles. + +in this case, the shared sides are essentially used as secondary validation checks after the primary star identification algorithm. they aren't actively used as part of determining star identities, but are instead used to confirm that identities are valid. the primary determination of star identities uses methods similar to earlier papers, based on non-iterative direct matching of accurate star pair angles and brightnesses. the 'shared side validation' is an additional operation added on top of that. + +